Legion laptop GPU thermal throttling

[Ryanull23]

Hello!

 

! Long post ahead

 

I have a Legion Y540 with an i5-9300h and RTX 2060, 2x 8GB RAM. Windows 11.

 

For a couple of days now I've had an issue with my GPU. It seems to throttle itself due to thermals. MSI Afterburner reports TEMPERATURE under limit overlay and the GPU underclocks itself down from 960 MHz (base) to 600-700, 400, it's all over the place. From what I've read online, it's rated for 80W, but power consumption only goes up to 47-53W under 99% load (as reported by MSI Afterburner). Max GPU temps seem to settle around 72-74*C. If I remember correctly the max temp until it should start to throttle would be 82-85C.

 

Since it's a laptop, i think there's something called BD Prochot, so if either the GPU or CPU hits its thermal limit, the other may underclock itself. I used ThrottleStop to disable that and undervolted the CPU to -165 mV core and -135 mV cache to keep it cooler. Max CPU temps while gaming are around 78-92*C and I set the thermal throttling limit to 99*C, CPU stays pinned at 4.0 GHz (set in ThrottleStop), with some games reporting 35-50% load (e.g. Control). I tried reflashing the VBIOS, i tried a different one too, with its TDP upped to 90W, but no success.

 

Obviously, the framerates drop a lot, down from maybe 80 to high 40s, 50s.

 

Looking up YT videos, I see that their GPU boosts up to 1300-1450 MHz and stays around there, whereas mine does this once then it clocks down. If I'm not mistaken, the boost clock for the GPU should be 1200 MHz, but as stated before, mine doesn't stay there, except for a second or two whiIe loading the game or alt tabbing back in. I tried a slight OC in MSI Afterburner, but the issue persists. Everything is set to High / Max Performance (Nvidia settings, power control settings, anything i could find). Legion control app (or whatever is called) thermal performance is set to Performance (balanced and quiet available). Hybrid mode (iGPU is disabled), discrete graphics usage is also enabled in the BIOS.

 

Voltage seems to be around 0.6 - 0.7v, but I do not know if it's normal, will need to research more. I cannot modify it in MSI Afterburner anyway. All i can control is Core and Memory clock speeds.

 

I used DDU to uninstall drivers, tried latest Nvidia driver as well as the one on the Lenovo website (performance was way worse with this one), but the issue persists.

 

Power brick looks fine, the laptop is receiving power, it's charging normally, i also tried disconnecting the battery and using it only with the AC adapter, no change. Fans are not dusty, i cleaned it, change thermal paste approx. 2 weeks ago.

 

I tried downgrading the BIOS, just to rule it out, been using the same version for the last 3 months or so (was on the latest version), but that failed and I'll hopefully get it fixed tomorrow.

 

Any ideas on what next troubleshooting step should be?

[Bombastinator]

Laptops of that era are famous for thermal throttling.  Not necessarily much to be done about it.  The CPUs were as hot as any seen in a long time.  A Standard first move is blow out the coolers as they can fill up with dust bunnies and stuff.  Next would be see if any of the fans are damaged.  They will probably need OEM replacements if they are. The only other thing I know to do is repaste the cooler with some high end TIM. Often buys a few degrees, but not a whole lot.   There could conceivably be other problems but they’re not likely.

[Ryanull23]

Fans are fine, not broken, connector is fine, i dusted them, they spin just fine. On YouTube, same laptop performs differently speaking of GPU clock speed. I don't have the possibility to check another one physically. Repasted two weeks ago or so, temps dropped from low 80s to low 70s

[Bombastinator]

4 hours ago, Ryanull23 said:

Fans are fine, not broken, connector is fine, i dusted them, they spin just fine. On YouTube, same laptop performs differently speaking of GPU clock speed. I don't have the possibility to check another one physically. Repasted two weeks ago or so, temps dropped from low 80s to low 70s

Hmm.  Would imply a software issue then except for the opening it up thing.  Did this slowness start happening when you opened the machine to repaste originally?

[Ryanull23]

8 hours ago, Bombastinator said:

Hmm.  Would imply a software issue then except for the opening it up thing.  Did this slowness start happening when you opened the machine to repaste originally?

Uhh, no. I usually repaste and dust it every 6 months or so. Temps seem to climb back up to the high 80s-90s on the CPU (pump out?). In the past 2 years (almost) i have repasted it probably 8 times or more, trying out different brands to see which seems to last the longest time before CPU temps start climbing, GPU was always mostly fine. It was always non conductive thermal paste and i steered clear of liquid metal.

 

Issue started happening probably a week or so ago, and i began to get annoyed by it few days ago when I tried out two games i hasn't played before. Repaste was 2, 2 and a half or maybe 3 weeks ago, can't really quite remember. At first i blamed my 20 tabs of Chrome and Twitch in the background then I decided to check it out because something felt odd still, and that's when I noticed the GPU clocks all over the place and what appears to be thermal throttling.

[Bombastinator]

2 hours ago, Ryanull23 said:

Uhh, no. I usually repaste and dust it every 6 months or so. Temps seem to climb back up to the high 80s-90s on the CPU (pump out?). In the past 2 years (almost) i have repasted it probably 8 times or more, trying out different brands to see which seems to last the longest time before CPU temps start climbing, GPU was always mostly fine. It was always non conductive thermal paste and i steered clear of liquid metal.

 

Issue started happening probably a week or so ago, and i began to get annoyed by it few days ago when I tried out two games i hasn't played before. Repaste was 2, 2 and a half or maybe 3 weeks ago, can't really quite remember. At first i blamed my 20 tabs of Chrome and Twitch in the background then I decided to check it out because something felt odd still, and that's when I noticed the GPU clocks all over the place and what appears to be thermal throttling.

Well that knocks that on the head.  There’s a time period though.  Anything else happen to the thing just before it started misbehaving?  Might not even have been something you did.  Could have been an auto update.

[Ryanull23]

Nothing I can think of. May have been an auto update, i don't know. I will backup data and do a format soon. See if that helps.

[Bombastinator]

6 hours ago, Ryanull23 said:

Nothing I can think of. May have been an auto update, i don't know. I will backup data and do a format soon. See if that helps.

Something has to have happened.  Computers are too dumb to do things by themselves.

[Ryanull23]

New update. GPU seems to have failed. I get no display out on the laptop screen or via HDMI or DisplayPort (tried a TV and a monitor). Seems weird to me that both integrated and discrete GPU would fail at the same time, but there it is. I left in in a game and came back 40 min later to black screen. It powers on, keyboard and mouse light up, i can hear the windows logon sound, but no display. I requested a RMA, it will be picked up Thursday. 

[Bombastinator]

2 hours ago, Ryanull23 said:

New update. GPU seems to have failed. I get no display out on the laptop screen or via HDMI or DisplayPort (tried a TV and a monitor). Seems weird to me that both integrated and discrete GPU would fail at the same time, but there it is. I left in in a game and came back 40 min later to black screen. It powers on, keyboard and mouse light up, i can hear the windows logon sound, but no display. I requested a RMA, it will be picked up Thursday. 

Hardware failure in a laptop usually means manufacturer repair.  Occasionally one can get lucky with an independent repair shop or DIY, but manufacturers also don’t often have much in the way of support or old parts, so not uncommonly they’re e-waste.

[Ryanull23]

1 hour ago, Bombastinator said:

Hardware failure in a laptop usually means manufacturer repair.  Occasionally one can get lucky with an independent repair shop or DIY, but manufacturers also don’t often have much in the way of support or old parts, so not uncommonly they’re e-waste.

Well, it's still under warranty. 2 months until it expires. Hopefully they'll fix / replace it.

 

The weird thing is I had 2 desktop GPUs fail in the past, but they both showed signs before, like driver randomly crashing, freezing, lines on the screen.

[Bombastinator]

30 minutes ago, Ryanull23 said:

Well, it's still under warranty. 2 months until it expires. Hopefully they'll fix / replace it.

 

The weird thing is I had 2 desktop GPUs fail in the past, but they both showed signs before, like driver randomly crashing, freezing, lines on the screen.

Well that’s a win.  If it’s hardware they’ll fix it, if it’s software they generally won’t.  Unless they can’t figure it out either in which case they can’t prove software or PEBKAC so they just give you a new one.  Can take a while though.

[Ryanull23]

On 12/1/2021 at 12:10 AM, Bombastinator said:

Well that’s a win.  If it’s hardware they’ll fix it, if it’s software they generally won’t.  Unless they can’t figure it out either in which case they can’t prove software or PEBKAC so they just give you a new one.  Can take a while though.

Okay, so the repair center called me (not Lenovo, 3rd party that works with the retailer where i bought it from) and they said they saw signs of liquid damage. $1000 for a new motherboard. I said i don't want to repair it then. I tried explaining that there's no way there was contact with any liquid. Its never been even close to a drop of water, they said they'd send pictures and my device back. 

 

Guess it's time for a new PC. Thanks for the assistance.

 

Edit: attached photos

Edited December 7, 2021 by Ryanull23
Attached pics

[Bombastinator]

Ouch.  Looks like they were digging hard. That one spot does look like liquid.  Some sort of soda.  Long dried.  Not much. Enough to maybe short some pins though. Looks like one drop in just the wrong place.  I don’t know if that is what actually did it or not.  Only a few pins were hit.  Few enough to follow.  It might be like the MacBook thing where they put a 50w lead near a lead that goes directly to the cpu.  Get anything on that connector and poof. 50w to the cpu. (Instead of around 1). It might not though and it’s being used to duck the repair.  I just don’t know.  There does appear to be a color shift difference. I don’t know if the third photo (the one where it is clear it is dried soda)  is on a brown strip not visible in the other pics.  Might just be the other side of the ribbon cable though. These are all pictures of one point of you look at the code numbers on the board.  I don’t know.

Edited December 7, 2021 by Bombastinator